Geospatial Susceptibility Assessment of Landslide in Battagram, Khyber Pakhtunkhwa, Pakistan

Abstract

A landslide is a natural disaster that can cause significant global damage  and human casualties. As a flood-prone area, the Battagram district of Khyber Pakhtunkhwa, Pakistan, has seen an increase in urbanization, making it challenging to choose an appropriate location for seismic activity. This study seeks to assess the susceptibility to landslide risk through the application such as seismic activity and flooding. This analysis employs Geographic Information System (GIS) and Remote Sensing techniques. The research utilized several data sets, encompassing geological data processed with the ArcGIS 10.8 software, Shuttle Radar Topography Mission (SRTM) data, Landsat thermal images from missions 5 and 8, thematic data, meteorological data, and a seismic catalogue. SAR photos are used to map Sentinel-1A in Google Earth Engine (GEE) to determine the extent of floods. The landslide inventory was separated into training and validation sets for this investigation. Significant contributing factors, including slope aspect, elevation, land cover and use during earthquakes, normalized difference vegetation index (NDVI), road distance, fault distance, rainfall, and geology, are taken into consideration when assessing landslip 
susceptibility. To establish the spatial correlation between landslides and these parameters, the frequency ratio model and weighted sum analysis were utilized. The WSM analysis indicates that 1.74% of the region is classified as having very low susceptibility, with the remaining areas being classified as low (14.26%), moderate (36.01%), high (2.57%), and very high (5.41%). 44.67% of the region is classified as having very high susceptibility by the FR model, with high (40.94%), moderate (11.61%), low (1.96%), and very low (0.79%) following. The FR model demonstrated reliability in risk assessment, with an accuracy of 85.7% against known landslide events. These findings support the use of GIS-based statistical modeling in urban planning and hazard mitigation by demonstrating how well it can identify high-risk areas. For increased accuracy and scalability, future developments should concentrate on adding more localized data.